I have enough data in my stats to take a stab at this question. Just looking at the stats some of my posts have been read a lot more than others and so just scanning the data it looks roughly like a power law, which Nate discusses a lot in his book. Here’s a shot at the raw data (#posts vs #reads/post):
Doesn’t look like much with linear axis, but as usual to see any pattern we need to look at a log-log plot:
Now at least the middle of the range looks fairly linear and thus matching a power law.
But, this really doesn’t address “rare” events (all the points show on the y-axis). There simply isn’t enough data to sense of these. And I can’t think of much (other than grouping observations in intervals, none of which is working very well), so I still don’t have anywhere near enough data for any analysis here.
Maybe in a year, but I also don’t think it’s going to work out then either, so we’ll see – yet another prediction to test in the future.